Chapter 8
A satellite-internet system is only useful once a terminal on a roof, a ship or an aircraft can register, find a satellite, hold a link through rain and Doppler, and move to the next satellite without dropping a TCP session. WIA-COMM-005 splits that work across four layers: data formats for constellations, inter-satellite links, ground stations and terminals; an API surface; handover, frequency-coordination, orbital and debris protocols; and a regulatory and integration layer. This chapter covers the last step: how an operator shows that a design built from those layers meets its targets, and what must be in place before service starts.
Three kinds of evidence matter. The first is radio evidence: link-budget margin and the behaviour of adaptive coding and modulation. The second is behavioural evidence: end-to-end round-trip time and handover interruption. The third is administrative evidence: spectrum filings, national licences and an end-of-life disposal plan. The sections below take each in turn, using the figures and procedures the standard defines.
The Phase 2 API gives integrators four endpoints. Together they cover the lifecycle of a user terminal, from first contact to ongoing link observation:
POST /satinternet/v1/terminal/register # register a user terminal
GET /satinternet/v1/coverage/{lat}/{lon} # coverage at a location
POST /satinternet/v1/handover/request # request a handover
GET /satinternet/v1/link/{id}/budget # link-budget telemetryA client finds the service through the discovery convention at /.well-known/wia-satellite-internet. The coverage endpoint takes a latitude and longitude and returns the list of visible satellites. Each entry carries a predicted elevation, a predicted link quality and the next visibility window. That output serves two consumers: the terminal's antenna-pointing logic, and operational dashboards that show where service is expected to be good or marginal.
The link-budget endpoint returns the quantities that decide whether a link closes: transmit power, antenna gain, free-space path loss, atmospheric loss, rain-fade margin and noise figure. For active links it is polled at high frequency, typically every minute, so operations teams can see a fading event as it develops. The downlink arithmetic behind those fields is:
P_rx = EIRP_sat - L_path - L_atm - L_rain + G_rx - L_rx
SNR = P_rx - N, N = kTBThe typical values the standard gives for a LEO Ka-band downlink are:
| Term | Meaning | Typical value |
|---|---|---|
| EIRP_sat | Satellite effective isotropic radiated power | 50-60 dBW |
| L_path | Free-space loss, 20 log₁₀(4πd/λ) | 165-175 dB |
| L_atm | Atmospheric absorption | 0.5-2 dB |
| L_rain | Rain fade, depends on frequency and climate | 1-10 dB |
| G_rx | User-terminal antenna gain | 35-40 dBi |
| L_rx | Receiver losses | 1-2 dB |
The same terms on the uplink use a user-terminal EIRP of 35-45 dBW and a satellite antenna gain of 25-30 dBi. The noise term N = kTB uses a system noise temperature of 100-300 K. Because the telemetry carries these components separately rather than a single "quality" number, an operator can tell whether a degraded link is losing margin to rain, to pointing error or to the receiver chain.
Validation starts from the margin the link budget leaves. The standard's target is 3-6 dB of link-budget margin, with rain fade and pointing error accounted for. Clear-sky SNR is typically 15-25 dB and falls to 5-15 dB in rain fade. Adaptive coding and modulation (ACM) is the mechanism that turns this swing into a graceful change in throughput instead of an outage. The reference ladder is:
| Condition | Modulation and code rate | Throughput |
|---|---|---|
| Clear sky | 32APSK, rate 9/10 | 500 Mbps |
| Light rain | 16APSK, rate 3/4 | 300 Mbps |
| Heavy rain | QPSK, rate 1/2 | 100 Mbps |
A rain-fade test therefore checks more than whether the link survives. It checks that the modem steps down the ladder, uses the modulation and LDPC code rates DVB-S2X allows, and recovers as the fade clears.
Latency is validated by measuring end-to-end RTT and comparing it with the decomposition in the standard. For a 550 km LEO the propagation component is small, 2 × (550 km / 300,000 km/s), about 3.7 ms. Ground processing (5-10 ms) and network routing (5-15 ms) make up the rest, for a total of 20-40 ms. Under that budget the pass criterion for LEO is an RTT below 40 ms. For orientation, MEO sits at 100-150 ms and GEO at 500-600 ms. If a LEO measurement lands in the GEO range, the problem is not the radio. Traffic is probably taking extra ground hops, which is the case inter-satellite routing is meant to avoid.
Doppler is the third radio-side check. At 28 GHz and 7.59 km/s orbital velocity, the maximum shift is:
Δf_max = (7590 / 299792) × 28 × 10⁹ ≈ ±710 kHz
df/dt ≈ 10-50 kHz/s (requires continuous tracking)A modem that only corrects Doppler at acquisition will lose lock within seconds. Validation should therefore confirm that frequency tracking runs continuously. It should also confirm that the correction is re-applied when the beam moves to a new satellite, because that is when the Doppler sign and rate change.
The orbit regimes in the standard map onto distinct commercial roles, and implementers usually pick a mix rather than a single regime. LEO constellations of the kind operated or planned by Starlink, OneWeb and Kuiper provide low-latency user access. MEO fills regional coverage and serves as a backup for LEO, and also carries the navigation constellations. GEO remains the economical choice for television broadcast and direct-to-home, weather monitoring and emergency backup, and it carries a legacy of maritime and aviation service. The standard's hybrid architecture combines the three: LEO for access, MEO for regional fill, GEO for broadcast and backup.
The cost structure differs by tier in ways that shape implementation. A LEO system needs a large number of satellites and a continuous handover engine. It also needs a terminal with a phased-array antenna that steers electronically, with no moving parts. A GEO system needs few satellites, but its 500-600 ms round trip rules out interactive uses that a LEO service can support. Ground infrastructure is a third cost driver. Gateways are sited near internet exchange points, with low latency to major data centres, in diverse geographies and in regulatory-friendly jurisdictions. A gateway placed far from an exchange point adds back the latency the constellation saved.
Terminals are the part of the market most exposed to physical limits. The standard lists a power draw of 20-50 W idle and 50-150 W active, with peaks of 100-200 W. That figure matters for anyone planning solar-powered rural sites, vessels or vehicles, where the power budget is as binding as the link budget.
The standard's stated purpose is universal access to remote, rural, maritime and aviation environments. Each puts a different stress on the implementation.
Mobility is what the handover protocol exists for. A LEO satellite is visible for only 4-7 minutes, so the terminal must change satellites constantly. The standard distinguishes three kinds of handover:
| Type | Frequency | Trigger |
|---|---|---|
| Intra-plane | Every 4-7 minutes | Satellite elevation below threshold; next satellite in the same plane |
| Inter-plane | Every 20-40 minutes | Better SNR available on a different plane |
| Gateway | Variable | Traffic optimisation; transparent to the user |
The protocol itself runs in three phases. About 30 seconds ahead the terminal measures the current SNR, scans for the next satellite and measures Doppler. About 10 seconds ahead it selects the best candidate, reserves resources and prepares beam steering. At execution it switches the beam, adjusts frequency for Doppler and keeps TCP connections alive without packet loss. The handover/request endpoint is how that exchange is invoked over the API.
Technical readiness has to run in parallel with a long administrative path. The ITU filing sequence in the standard sets the pace:
| Step | Timing |
|---|---|
| Advance publication | 7 years before launch |
| Coordination request | 5 years before launch |
| Notification | Within 30 days of satellite operation |
| Frequency assignment | Recorded in the ITU Master Register |
The underlying ITU Radio Regulations articles are Article 9 (frequency coordination), Article 11 (notification and recording) and Article 22 (operational procedures), plus the EIRP limits referenced in RR Appendix 4. National licences sit on top. In the United States the FCC process uses IBFS and a processing-round system for NGSO constellations, with milestones such as launching 50% of the constellation within six years. In the United Kingdom the route is an Ofcom spectrum access licence plus coordination with existing services. Operators in other jurisdictions (Japan's MIC, Korea's KCC, EU member-state regulators) face their own requirements. For that reason the discovery document declares the operator's licensed jurisdictions, and each data envelope carries a jurisdiction identifier, so compliance can be traced region by region.
Coordination with other spectrum users is also tested in practice. The measures in the standard are orthogonal polarisations, spot beams with frequency reuse, real-time interference monitoring with adaptive power control, and protection of the GEO arc through off-axis EIRP limits and exclusion zones of ±2° around the arc. Export control is a separate track. Satellite technology falls under ITAR in the United States, which requires export licences with exceptions for commercial systems, and under the multilateral Wassenaar Arrangement.
Debris mitigation is a compliance obligation, not an end-of-life afterthought. LEO satellites must deorbit within 25 years of end of mission. Active deorbit uses remaining propellant for controlled reentry in 1-12 months. Passive decay below 600 km relies on drag over 1-5 years. A GEO satellite is raised at least 300 km out of the GEO belt. In operation, daily conjunction screening against the TLE catalogue raises an alert when collision probability exceeds 10⁻⁴, and a manoeuvre is triggered when the predicted miss distance is under 1 km. Design for demise favours aluminium and composites over high-melting-point titanium or inconel, keeps surviving components under about 5 kg each, and depletes propellant tanks before deorbit to prevent on-orbit explosions. ISO 24113 is the governing requirements document.
The standard also ships a public conformance test suite that exercises every endpoint and protocol exchange. A reference container, wia/satellite-internet-host:1.0.0, serves every Phase 2 endpoint with mock data, so integrators can code against it before touching a live network.
The standard's design checklist is effectively the build order for a new system:
[ ] Define orbit parameters (altitude, inclination, planes)
[ ] Calculate constellation size for desired coverage
[ ] Design link budget (uplink, downlink, ISL)
[ ] Select frequency bands and file with ITU
[ ] Design ground station network
[ ] Develop user terminal (phased array or parabolic)
[ ] Implement handover algorithms
[ ] Plan debris mitigation and deorbit strategy
[ ] Obtain regulatory approvals (FCC, ITU, national)
[ ] Manufacture, test, and launch satellitesThe order has a logic. Orbit and constellation size come first, because minimum elevation angle, coverage radius and phasing all follow from altitude. For a minimum elevation of 25-40°, the coverage radius is r = R_e × arccos[R_e × cos(θ) / (R_e + h)]. Frequency selection follows, because the band fixes both the filing path and the rain-fade exposure. Ka-band (27.5-30.0 GHz up, 17.7-20.2 GHz down) is the primary choice for LEO, Ku-band is the legacy GEO choice, and V-band is the future high-throughput option. Filing starts early because the lead times are measured in years, while hardware lead times are measured in months. Handover algorithms and debris plans are not post-launch details. They are required evidence for the filings and for terminal behaviour on day one.
A first deployment usually targets a single region, with a ground-station network and an initial allocation of user terminals. The reference deployment guide covers per-region capacity sizing, the cadence of spectrum coordination and an operations runbook for the first 90 days of service. During those 90 days the link-budget telemetry earns its keep. A rain-fade profile that differs from the design assumption shows up as repeated ACM step-downs, and an under-margined pointing algorithm shows up as terminals that sit chronically low on the ladder.
Acceptance testing in the standard has four parts:
Several common mistakes follow from the physics. Testing only in clear sky hides ACM problems. Measuring latency to a nearby server hides the cost of a poorly placed gateway. Counting a handover as seamless because the session survived hides interruptions long enough to damage real-time applications. Protocol choices matter too: the standard names TCP BBR, QUIC and custom congestion control as ways to avoid long-RTT inefficiencies on a path whose delay changes at every handover.
Integration with the wider WIA family is optional but defined. WIA-INTENT lets a customer request something like 100 Mbps, under 50 ms latency and 99.9% uptime, and have the network configured to match. WIA-OMNI-API provides subscriber identity and a unified interface across operators. WIA-AIR-SHIELD supplies a trust list for coordination between operators, and WIA-INFRA-MONITORING consumes ground-station and satellite telemetry.
The standard's own roadmap is incremental. Version 1.0.0 fixes the LEO, MEO and GEO baseline. The 1.1.x line adds constellation tiers and standardisation of optical inter-satellite links, and 1.2.x adds confidential satellite-internet processing inside trusted execution environments. A possible 2.0.0 would introduce post-quantum signatures, which would be a breaking change.
Optical links are the technically significant item. At 1550 nm, with data rates of 1-100 Gbps per link, ranges up to 5,000 km and beam divergence under 10 microradians, they offer 10-100 times the bandwidth of RF. They need no frequency licence and are hard to intercept. Standardising them would make cross-operator mesh routing practical, where BGP already handles inter-constellation routing and OSPF-style routing handles the inside. Delay-tolerant networking (RFC 4838) covers the deep-space case. Alongside optical links, V-band (37.5-42.5 GHz down, 47.2-50.2 GHz up, up to 10 GHz total) promises more capacity, but it will raise the rain-fade problem the ACM ladder already addresses. Convergence with 5G non-terrestrial networks, tracked through 3GPP TS 23.501 and the NTN specifications, will bring satellite terminals closer to the mobile-network model.
This chapter treated certification as an engineering sequence, not a paperwork exercise. Four API endpoints and a discovery path let a terminal register, look up coverage, request handover and report link-budget telemetry. The budget equation, the ACM ladder and the 3-6 dB margin target turn that telemetry into a pass-or-fail judgement. The RTT decomposition (3.7 ms propagation inside a 20-40 ms total), the continuous Doppler tracking of about ±710 kHz at 28 GHz, and the 100 ms handover interruption limit define how the system should behave. The regulatory path, from ITU advance publication seven years before launch through national licences, export control and a 25-year deorbit commitment, runs in parallel and is usually the longer one.
The practical lesson is that an operator proves a satellite-internet system by measuring it under fade, motion and handover, and by holding filings that match what the system actually does. The next chapter moves from one operator to many. It looks at what changes when two constellations must share spectrum, exchange trust and federate service, building on the jurisdiction identifiers, trust lists and telemetry established here.